1. Field of the Invention
The present invention relates in particular to a cluster monitor and a method for monitoring a cluster for activating a monitoring agent in a high-availability cluster system, and a computer-readable recording medium in which a program for implementing the same is recorded.
2. Background Art
Conventionally, agent-based monitoring is performed to monitor business applications such as databases. In the agent-based monitoring, a monitoring program called “monitoring agent” is connected to a server that executes a business application, in the same manner as access from a client. The monitoring agent makes a dummy request to the business application, and checks whether or not a normal response is given from the business application (e.g., see JP 2008-293117A).
Specifically, JP 2008-293117A discloses a system in which a monitoring agent is constructed on a computer other than a server being a target of monitoring. Unlike the case where system components such as a server, a disk, and a network are separately monitored, with the system disclosed in JP 2008-293117A the monitoring agent is able to perform monitoring from a viewpoint on whether or not an appropriate service is being provided, standing in the position of system users.
In recent years, HA (High-Availability) cluster systems are available. An HA cluster system is a system that causes a number of servers to operate as if a single server operates. In this system, even if one of the servers stops due to a failure or the like, processing can be continued without stopping the entire system, while the stopped server can be repaired and replaced. Furthermore, the system disclosed in JP 2008-293117A is considered to be applicable to such an HA cluster system.
Meanwhile, in the system disclosed in JP 2008-293117A mentioned above, the server on which the monitoring agent is activated is fixed, and it is impossible to dynamically determine which server is most appropriate for activation of the monitoring agent and activate the monitoring agent on this server. Therefore, in the case where the system disclosed in JP 2008-293117A is applied to an HA cluster system, monitoring cannot be performed after the server on which the monitoring agent is activated goes down.
In addition, in the system disclosed in JP 2008-293117A the monitoring agent cannot be activated on the server on which the business application is being executed, which is required in some cases from the viewpoint of load distribution. If the system disclosed in JP 2008-293117A is applied to an HA cluster system, a problem arises in that the load cannot be sufficiently distributed.